Environmental Products
Land cover characteristics key to watershed restoration and other applications can be effectively derived from commercial spectral imagery using a combination of traditional and advanced AAI-developed techniques. Automatic land cover identification and land use classification, for example, provide a practical means for inventorying the full watershed. AAI's Mixed Material Classifier technology provides effective foliage penetration, particularly valuable along river banks and at suspected stressor sites. Species-level mapping of native and invasive vegetation with AAI's iCee™ Atmospheric Correction, Mixed Material Classifier, and Material Identifier provide an efficient means for locating and monitoring changes in species-level acreage over time.Changes in land cover and land use over time at site-level to full landscape scales can also be quantitatively characterized using these technologies. Searches using these technologies for material disturbances and anomalies, many at sub-visual scales, can further identify specific stressor sources or other subtle features that may otherwise be missed. Restoration targets for impaired land areas can be established using imagery from an earlier designated baseline year. Progress against the targets can then be quantitatively assessed using a periodic monitoring program.
Automatic Land Cover Identification and Water Segmentation
Automatic water - land cover segmentation and identification of an area on Cape Cod, MA (above).
Coastal and Riverine Shoreline Characterization
The area below is a zoomed-in portion of the image above, showing detections of subpixel-scale shoreline debris in red. The debris corresponds to specific water levels, e.g., mean high water, high high water, etc, which can be used for riverine and coastal shoreline water level delineation.
The shoreline debris can be used to estimate riverine and coastal shoreline slopes and gradients (above).
Mixed Material Classifier can be used to "penetrate" the tree canopy to identify forest floor materials. Mixed Material Classifier identifies up to three component materials per pixel in an image, allowing subpixel identification of forest floor materials exposed through canopy "holes" (above right). Locations where sand and silt meet the vegetated understory (below, shown in red), for example, represent the under-canopy high water boundary.
Watershed Surveillance
Watershed inventories, like the one shown below for Ft Cobb, OK, can be performed
at both local site-level and broad area landscape-level scales.
Species-Level Mapping and Wetlands Delineation
AAI's subpixel analysis technologies, such as Mixed Material Classifer, can perform species-level mapping of wetland vegetation species in watersheds. In the example below, subpixel analysis processing identified tupelo trees (yellow) and cypress trees (blue) in this forest environment, almost all in "mixed" pixels with other species. Each tree species thrives in wetland areas, although tupelo is not obligate. The presence of both species together (green areas) indicates a high probability of wetlands. A ground truth field verification study (NASA, US Forest Service, and Clemson University) revealed identification accuracies near 90% (omission and commission) for both tree species.
Land Cover Change Analysis
AAI's Material Identifier can be used with geo-registered multi-date imagery to characterize the land cover on the two dates (below left). Note that the forest land cover (dark green) reduced dramatically in this Oklahoma watershed between 1985 and 2001. Changes in land-cover between the two dates can be automatically characterized (below right, top). Using the USGS river/stream overlay, changes within the 100m riparian buffer zones can be identified and quantified by area (below right, bottom). Pastures in the riparian buffer zones are likely sources of water pollution, and there was substantial forest-to-pasture land-cover change from 1985 to 2001 (red areas).
Using a similar quantitative change analysis approach to the one above, changes in percentage tree cover over a 24-year period were assessed for the Seattle/Tacoma metro area (below). Areas with over 50% tree cover are in various shades of green. Areas with 20-50% tree cover are in shades of brown, while black indicates urban areas with less than 20% tree cover. Quantitative tree cover acreage data (not shown) revealed dramatic ecosystem changes resulting from urbanization over that period.
Quantitative assessments of the extent of clear-cutting in South America for planting illicit crops have been enabled by AAI's subpixel analysis technologies (see Mixed Material Classifier). In the illustration below, subpixel analysis processing accurately identified and quantified clear-cut areas where most occupied less than a whole pixel (yellow and orange areas in right image). Traditional whole-pixel spectral classifiers have difficulty identifying small areas that occupy less than a whole pixel.
Impervious Surface Mapping
Land cover and impervious surface mapping can be effectively combined for storm-water management applications, such as the one shown below for the City of New Bern, N.C. In this case an IKONOS satellite image was used (below left). The subpixel-scale impervious surface map accurately quantified the area covered by roads, parking lots, driveways, sidewalks, roofs, tennis courts, etc. This, combined with the six requested general land cover classes (below right, top) provided the data needed for a storm water management assessment.
Chemical Spill Contamination Sites
The ability to detect subtle subpixel material components in mixed pixels enabled the successful search for residual "stains" associated with prior chemical spills, with the goal of identifying potential contamination sites on the host military base and associated watershed. A chemical spill had recently occurred within the DRMO area (shown as Site 2 in the image below), and the clean-up excavation material was temporarily deposited in the "Piles Area" (shown as Site 1 in the image below) for de-volatilization. The piles were temporarily uncovered at the time of satellite overpass, and subpixel spectral signatures of the stained excavation soil were successfully extracted from image pixels at Site 1. AAI's Signature Adapter was used to develop an adaptive signature to enable detection and discrimination of the stains on any unknown host material. Detections of residual chemical spill "stains" were detected at 7 sites, the DRMO area, piles area, and five others. Ground truth field inspections by military personnel confirmed two jet fuel spills on the airbase, one on the tarmac (Site 4) and one on a runway (Site 6). A fuel spill was also confirmed at a marine repair facility (Site 7). The other two sites have not yet been assessed.
Environmental Products List